Recent searches

No recent searches

Search options

Only available when logged in.
mastodon.world is one of the many independent Mastodon servers you can use to participate in the fediverse.
Generic Mastodon server for anyone to use.

Administered by:

Server stats:

8.1K
active users

mastodon.world: AboutStatusProfiles directoryPrivacy policy

Mastodon: AboutGet the appKeyboard shortcutsView source codev4.4.0

Sarah Lea

To learn from experience, a reinforcement learning (RL) agent needs four key elements:

:blobcoffee: State: What situation is the agent in?
:blobcoffee: Actions: What are the possible moves from here?
:blobcoffee: Reward: What does the agent receive after an action?
:blobcoffee: Value function: How good is a state (or action), both now and in the future?

This is the foundation of how an RL agent learns to make better decisions over time.

#reinforcementlearning#machinelearning#ai
Sarah Lea<p>Can you remember learning to walk as a baby? You didn’t read a manual. Neither does an AI agent.</p><p>Reinforcement Learning (RL) isn’t about knowing the correct answer.<br>It’s about learning through trial and error, by interacting with an environment &amp; receiving feedback.</p><p>That’s how AlphaGo defeated a world champion:<br>It first learned from expert games. Then it played against itself, millions of times, using RL to get better with each game. That’s how it mastered Go.</p><p><a href="https://techhub.social/tags/machinelearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>machinelearning</span></a> <a href="https://techhub.social/tags/ai" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ai</span></a> <a href="https://techhub.social/tags/ki" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ki</span></a> <a href="https://techhub.social/tags/google" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>google</span></a> <a href="https://techhub.social/tags/reinforcementlearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>reinforcementlearning</span></a> <a href="https://techhub.social/tags/alphago" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>alphago</span></a> <a href="https://techhub.social/tags/datascience" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>datascience</span></a> <a href="https://techhub.social/tags/datascientist" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>datascientist</span></a></p>
Sarah Lea<p>Which strategy do you use when learning something new?</p><p>3 strategies AI agents use to learn what works:<br>:blobcoffee: Greedy: Stick with what has worked best so far.<br>:blobcoffee: ε-Greedy: Mostly stick with the best. But try something new every now and then.<br>:blobcoffee: Optimistic Start: Assume everything is great until proven otherwise.</p><p>They all come from something called the “Multi-Armed Bandit” problem.</p><p>But they show up in real life too:<br>→ Trying a new café.<br>→ Deciding what to study <br>→ Choosing which project to pursue at work.</p><p>Which one do you use most often?<br>And should you change it?</p><p>Curious to dive deeper? I covered both topics in my latest two articles: <a href="https://towardsdatascience.com/author/schuerch_sarah/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">towardsdatascience.com/author/</span><span class="invisible">schuerch_sarah/</span></a></p><p><a href="https://techhub.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://techhub.social/tags/KI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>KI</span></a> <a href="https://techhub.social/tags/machinelearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>machinelearning</span></a> <a href="https://techhub.social/tags/reinforcementlearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>reinforcementlearning</span></a> <a href="https://techhub.social/tags/learning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>learning</span></a> <a href="https://techhub.social/tags/datascience" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>datascience</span></a></p>
michabbb<p><a href="https://social.vivaldi.net/tags/ART" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ART</span></a> Agent Reinforcement Trainer: <a href="https://social.vivaldi.net/tags/Opensource" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Opensource</span></a> <a href="https://social.vivaldi.net/tags/RL" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>RL</span></a> <a href="https://social.vivaldi.net/tags/framework" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>framework</span></a> for building reliable <a href="https://social.vivaldi.net/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> agents 🤖</p><p>🎯 Improved email agent success rate from 74% to 94% using <a href="https://social.vivaldi.net/tags/ReinforcementLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ReinforcementLearning</span></a> on <a href="https://social.vivaldi.net/tags/Qwen" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Qwen</span></a> model 💰 Reduced costs from $55 to $0.80 per 1,000 requests 🧵👇</p>
Sarah Lea<p>🍕 Imagine trying two pizzerias and always going back to the one that seemed better. Sounds simple? Maybe.</p><p>But what if there’s another one in the city that’s even better – and you never tried it?</p><p>That’s what the greedy strategy does. It sticks to what worked best so far.</p><p>In Multi-Armed Bandits, a classic problem that helps us understand Reinforcement Learning, there are three common strategies to deal with this dilemma:<br>:blobcoffee: Greedy<br>:blobcoffee: ε-Greedy<br>:blobcoffee: Optimistic Initial Values</p><p><a href="https://towardsdatascience.com/simple-guide-to-multi-armed-bandits-a-key-concept-before-reinforcement-learning/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">towardsdatascience.com/simple-</span><span class="invisible">guide-to-multi-armed-bandits-a-key-concept-before-reinforcement-learning/</span></a></p><p><a href="https://techhub.social/tags/DataScientist" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DataScientist</span></a> <a href="https://techhub.social/tags/datascience" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>datascience</span></a> <a href="https://techhub.social/tags/data" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>data</span></a> <a href="https://techhub.social/tags/machinelearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>machinelearning</span></a> <a href="https://techhub.social/tags/ai" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ai</span></a> <a href="https://techhub.social/tags/ki" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ki</span></a> <a href="https://techhub.social/tags/reinforcementlearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>reinforcementlearning</span></a> <a href="https://techhub.social/tags/agents" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>agents</span></a> <a href="https://techhub.social/tags/agenticai" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>agenticai</span></a> <a href="https://techhub.social/tags/technology" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>technology</span></a> <a href="https://techhub.social/tags/tech" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>tech</span></a></p>
Sarah Lea<p>Reinforcement Learning starts with a simple but powerful idea:<br>Trial &amp; Error. Learning what works.</p><p>The Multi-Armed Bandit problem is a first step into this world.<br>It's not just about slot machines. Iit's about how AI (and humans) learn to choose.</p><p><a href="https://towardsdatascience.com/simple-guide-to-multi-armed-bandits-a-key-concept-before-reinforcement-learning/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">towardsdatascience.com/simple-</span><span class="invisible">guide-to-multi-armed-bandits-a-key-concept-before-reinforcement-learning/</span></a></p><p><a href="https://techhub.social/tags/ReinforcementLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ReinforcementLearning</span></a> <a href="https://techhub.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://techhub.social/tags/CognitiveScience" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>CognitiveScience</span></a> <a href="https://techhub.social/tags/Psychology" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Psychology</span></a> <a href="https://techhub.social/tags/Behavior" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Behavior</span></a> <a href="https://techhub.social/tags/DecisionMaking" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DecisionMaking</span></a> <a href="https://techhub.social/tags/Bandits" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Bandits</span></a> <a href="https://techhub.social/tags/machinelearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>machinelearning</span></a> <a href="https://techhub.social/tags/KI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>KI</span></a> <a href="https://techhub.social/tags/Datascience" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Datascience</span></a> <a href="https://techhub.social/tags/datascientist" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>datascientist</span></a></p>
Sarah Lea<p>Do you always go to the same café? Or do you try something new?</p><p>That’s the exploration vs. exploitation dilemma: Decision under uncertainty.</p><p>Multi-armed bandits model exactly that.</p><p>And this dilemma shows up everywhere: Recommender systems, A/B tests, online ads, even in human psychology.</p><p>Nobel Prize winner Daniel Kahneman called this one of the most fundamental cognitive patterns.</p><p>🎰 I explain what it is, why it matters, and how AI systems handle it. </p><p>:blobcoffee: Full article here: <a href="https://towardsdatascience.com/simple-guide-to-multi-armed-bandits-a-key-concept-before-reinforcement-learning/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">towardsdatascience.com/simple-</span><span class="invisible">guide-to-multi-armed-bandits-a-key-concept-before-reinforcement-learning/</span></a></p><p><a href="https://techhub.social/tags/ReinforcementLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ReinforcementLearning</span></a> <a href="https://techhub.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://techhub.social/tags/CognitiveScience" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>CognitiveScience</span></a> <a href="https://techhub.social/tags/Kahneman" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Kahneman</span></a> <a href="https://techhub.social/tags/Psychology" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Psychology</span></a> <a href="https://techhub.social/tags/Behavior" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Behavior</span></a> <a href="https://techhub.social/tags/DecisionMaking" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DecisionMaking</span></a> <a href="https://techhub.social/tags/Bandits" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Bandits</span></a> <a href="https://techhub.social/tags/machinelearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>machinelearning</span></a> <a href="https://techhub.social/tags/KI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>KI</span></a> <a href="https://techhub.social/tags/Datascience" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Datascience</span></a> <a href="https://techhub.social/tags/datascientist" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>datascientist</span></a></p>
Hacker News<p>Shoggoth Mini – A soft tentacle robot powered by GPT-4o and RL</p><p><a href="https://www.matthieulc.com/posts/shoggoth-mini" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">matthieulc.com/posts/shoggoth-</span><span class="invisible">mini</span></a></p><p><a href="https://mastodon.social/tags/HackerNews" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>HackerNews</span></a> <a href="https://mastodon.social/tags/ShoggothMini" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ShoggothMini</span></a> <a href="https://mastodon.social/tags/SoftRobot" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>SoftRobot</span></a> <a href="https://mastodon.social/tags/GPT4o" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>GPT4o</span></a> <a href="https://mastodon.social/tags/ReinforcementLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ReinforcementLearning</span></a> <a href="https://mastodon.social/tags/TechInnovation" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>TechInnovation</span></a></p>
hlfshell<p>I had an idea scratching at the back of my head based on some of my recent talks given on AI research.</p><p>It's essentially hoping to train better generalist long-horizon robotic policies via a generative reward model and multi-turn credit assignment.</p><p><a href="https://hlfshell.ai/posts/proposal-robotic-complex-task-rl-training/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">hlfshell.ai/posts/proposal-rob</span><span class="invisible">otic-complex-task-rl-training/</span></a></p><p><a href="https://hachyderm.io/tags/research" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>research</span></a> <a href="https://hachyderm.io/tags/ai" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ai</span></a> <a href="https://hachyderm.io/tags/robotics" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>robotics</span></a> <a href="https://hachyderm.io/tags/reinforcementlearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>reinforcementlearning</span></a></p>
Hacker News<p>How to scale RL to 10^26 FLOPs</p><p><a href="https://blog.jxmo.io/p/how-to-scale-rl-to-1026-flops" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">blog.jxmo.io/p/how-to-scale-rl</span><span class="invisible">-to-1026-flops</span></a></p><p><a href="https://mastodon.social/tags/HackerNews" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>HackerNews</span></a> <a href="https://mastodon.social/tags/scaleRL" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>scaleRL</span></a> <a href="https://mastodon.social/tags/FLOPs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>FLOPs</span></a> <a href="https://mastodon.social/tags/reinforcementLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>reinforcementLearning</span></a> <a href="https://mastodon.social/tags/AIresearch" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AIresearch</span></a> <a href="https://mastodon.social/tags/optimization" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>optimization</span></a></p>
Hacker News<p>The upcoming GPT-3 moment for RL</p><p><a href="https://www.mechanize.work/blog/the-upcoming-gpt-3-moment-for-rl/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">mechanize.work/blog/the-upcomi</span><span class="invisible">ng-gpt-3-moment-for-rl/</span></a></p><p><a href="https://mastodon.social/tags/HackerNews" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>HackerNews</span></a> <a href="https://mastodon.social/tags/GPT3" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>GPT3</span></a> <a href="https://mastodon.social/tags/RL" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>RL</span></a> <a href="https://mastodon.social/tags/ReinforcementLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ReinforcementLearning</span></a> <a href="https://mastodon.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://mastodon.social/tags/Innovation" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Innovation</span></a></p>
hlfshell<p>The idea of using the multi-turn credit assignment combined with a more fleixble verifier system - like Deepseek's Generative Reward Model - creats a potentially FASCINATING idea that I wish I had the time and compute to explore.</p><p>Imagine a robotic system with a system dedicated to judging it (in simulation) based on a set of small piece meal sub-tasks for any long running tasks. You could in theory create a more robust generalzied model from this.</p><p><a href="https://hachyderm.io/tags/reinforcementlearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>reinforcementlearning</span></a> <a href="https://hachyderm.io/tags/robotics" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>robotics</span></a> <a href="https://hachyderm.io/tags/ai" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ai</span></a></p>
Hacker News<p>RULER – Easily apply RL to any agent</p><p><a href="https://openpipe.ai/blog/ruler" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">openpipe.ai/blog/ruler</span><span class="invisible"></span></a></p><p><a href="https://mastodon.social/tags/HackerNews" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>HackerNews</span></a> <a href="https://mastodon.social/tags/RULER" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>RULER</span></a> <a href="https://mastodon.social/tags/RL" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>RL</span></a> <a href="https://mastodon.social/tags/agents" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>agents</span></a> <a href="https://mastodon.social/tags/ReinforcementLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ReinforcementLearning</span></a> <a href="https://mastodon.social/tags/OpenAI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>OpenAI</span></a> <a href="https://mastodon.social/tags/AItools" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AItools</span></a></p>
LLMsA Professional Guide to Reinforcement Learning Models in Machine Learning By Moustafa Mohamed AI Developer | Machine Learning, Deep Learning, LLM Engineering LinkedIn | GitHub | Portfolio | Kaggle ...<br><br><a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/data-science" target="_blank">#data-science</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/reinforcement-learning" target="_blank">#reinforcement-learning</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/machine-learning" target="_blank">#machine-learning</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/ai" target="_blank">#ai</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/deep-learning" target="_blank">#deep-learning</a><br><br><a href="https://moustafamohamed01.medium.com/a-professional-guide-to-reinforcement-learning-models-in-machine-learning-2197bfdf5e5d?source=rss------machine_learning-5" rel="nofollow noopener" target="_blank">Origin</a> | <a href="https://awakari.com/sub-details.html?id=LLMs" rel="nofollow noopener" target="_blank">Interest</a> | <a href="https://awakari.com/pub-msg.html?id=N5ITRf6rnm6lHoZm21YAd5tWUCG&amp;interestId=LLMs" rel="nofollow noopener" target="_blank">Match</a>
Georg Weissenbacher<p>A nice <a href="https://fediscience.org/tags/arstechnica" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>arstechnica</span></a> article on <a href="https://fediscience.org/tags/ReinforcementLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ReinforcementLearning</span></a> in <a href="https://fediscience.org/tags/LLMs" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>LLMs</span></a> </p><p><a href="https://arstechnica.com/ai/2025/07/how-a-big-shift-in-training-llms-led-to-a-capability-explosion/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">arstechnica.com/ai/2025/07/how</span><span class="invisible">-a-big-shift-in-training-llms-led-to-a-capability-explosion/</span></a></p>
Ars Technica News<p>How a big shift in training LLMs led to a capability explosion <a href="https://arstechni.ca/FerG" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">arstechni.ca/FerG</span><span class="invisible"></span></a> <a href="https://c.im/tags/reinforcementlearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>reinforcementlearning</span></a> <a href="https://c.im/tags/imitationlearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>imitationlearning</span></a> <a href="https://c.im/tags/explainers" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>explainers</span></a> <a href="https://c.im/tags/Features" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Features</span></a> <a href="https://c.im/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a></p>
Erik Jonker<p>Good article how reinforcement learning improved current AI models. Also illustrates that LLMs today are not just imitating.<br><a href="https://arstechnica.com/ai/2025/07/how-a-big-shift-in-training-llms-led-to-a-capability-explosion/?utm_brand=arstechnica&amp;utm_social-type=owned&amp;utm_source=mastodon&amp;utm_medium=social" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">arstechnica.com/ai/2025/07/how</span><span class="invisible">-a-big-shift-in-training-llms-led-to-a-capability-explosion/?utm_brand=arstechnica&amp;utm_social-type=owned&amp;utm_source=mastodon&amp;utm_medium=social</span></a><br><a href="https://mastodon.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://mastodon.social/tags/reinforcementlearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>reinforcementlearning</span></a></p>
Hacker News<p>Reinforcement Learning from Human Feedback (RLHF) in Notebooks</p><p><a href="https://github.com/ash80/RLHF_in_notebooks" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">github.com/ash80/RLHF_in_noteb</span><span class="invisible">ooks</span></a></p><p><a href="https://mastodon.social/tags/HackerNews" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>HackerNews</span></a> <a href="https://mastodon.social/tags/ReinforcementLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ReinforcementLearning</span></a> <a href="https://mastodon.social/tags/HumanFeedback" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>HumanFeedback</span></a> <a href="https://mastodon.social/tags/RLHF" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>RLHF</span></a> <a href="https://mastodon.social/tags/Notebooks" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Notebooks</span></a> <a href="https://mastodon.social/tags/AIResearch" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AIResearch</span></a></p>
LLMsProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language Models The key idea The key idea Reinforcement learning (RL) has had a resurgence in LLMs with application to ...<br><br><a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/LLMs" target="_blank">#LLMs</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/training-dynamics" target="_blank">#training-dynamics</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/fine-tuning" target="_blank">#fine-tuning</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/reasoning" target="_blank">#reasoning</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/reinforcement-learning" target="_blank">#reinforcement-learning</a><br><br><a href="https://graphcore-research.github.io/prorl/" rel="nofollow noopener" target="_blank">Origin</a> | <a href="https://awakari.com/sub-details.html?id=LLMs" rel="nofollow noopener" target="_blank">Interest</a> | <a href="https://awakari.com/pub-msg.html?id=9N5KR0r2wXZVbgWCE3Sv9k3kf6e&amp;interestId=LLMs" rel="nofollow noopener" target="_blank">Match</a>
US<p><a href="https://www.europesays.com/us/41491/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="">europesays.com/us/41491/</span><span class="invisible"></span></a> Multi objective reinforcement learning driven task offloading algorithm for satellite edge computing networks <a href="https://pubeurope.com/tags/AerospaceEngineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AerospaceEngineering</span></a> <a href="https://pubeurope.com/tags/Computing" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Computing</span></a> <a href="https://pubeurope.com/tags/DQN" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DQN</span></a> <a href="https://pubeurope.com/tags/Engineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Engineering</span></a> <a href="https://pubeurope.com/tags/HumanitiesAndSocialSciences" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>HumanitiesAndSocialSciences</span></a> <a href="https://pubeurope.com/tags/multidisciplinary" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>multidisciplinary</span></a> <a href="https://pubeurope.com/tags/ReinforcementLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ReinforcementLearning</span></a> <a href="https://pubeurope.com/tags/SatelliteEdgeComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>SatelliteEdgeComputing</span></a> <a href="https://pubeurope.com/tags/Science" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Science</span></a> <a href="https://pubeurope.com/tags/SpaceBasedInformationNetwork" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>SpaceBasedInformationNetwork</span></a> <a href="https://pubeurope.com/tags/TaskOffloading" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>TaskOffloading</span></a> <a href="https://pubeurope.com/tags/Technology" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Technology</span></a> <a href="https://pubeurope.com/tags/UnitedStates" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>UnitedStates</span></a> <a href="https://pubeurope.com/tags/UnitedStates" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>UnitedStates</span></a> <a href="https://pubeurope.com/tags/US" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>US</span></a></p>
TrendingLive feeds
About
404Not Found